Activation Functions in Artificial Neural Networks: A Systematic Overview
It addresses confusion in theory and practice for those studying or applying neural networks, but is incremental as it synthesizes existing knowledge.
The paper tackles the proliferation of new activation functions in deep learning by providing an analytic and up-to-date overview of popular activation functions and their properties, serving as a timely resource for researchers and practitioners.
Activation functions shape the outputs of artificial neurons and, therefore, are integral parts of neural networks in general and deep learning in particular. Some activation functions, such as logistic and relu, have been used for many decades. But with deep learning becoming a mainstream research topic, new activation functions have mushroomed, leading to confusion in both theory and practice. This paper provides an analytic yet up-to-date overview of popular activation functions and their properties, which makes it a timely resource for anyone who studies or applies neural networks.